Tightly Coupled GPS/INS/CCD Integration Based on GPS Carrier Phase Velocity Update

摘要

In the majority of multi-sensor mapping systems, the GPS/INS integration serves as a navigation and georegistration tool, supporting the direct orientation of the simultaneously acquired imagery. In a tightly coupled integration architecture, where double difference (DD) GPS carrier phase observations and the IMU (inertial measurement unit) accelerometer and gyroscope data are integrated in a single centralized Kalman filter, any loss of GPS signal causes a discontinuity in the filter measurement update until the signal reacquisition and the integer ambiguity recovery. Long GPS gaps may cause IMU error growth, and thus, positioning error increase, which may prevent instantaneous ambiguity recovery, especially for longer baselines and noisy environments. If the time needed to fix the new ambiguities is excessive, preventing the filter from using any GPS information, it may be beneficial to switch to either triple differences (they do not require ambiguity fixing, as ambiguities are eliminated by the third difference), or to velocities derived from the carrier phase observations. Extracting the GPS velocity from the carrier phase observation, which is independent of the ambiguity value, offers an opportunity to calibrate the IMU sensors, and to maintain the high quality navigation parameter estimation. This paper presents a filter update method based on GPS velocity calibration in the GPS/INS integration system, to assure availability and continuity of the navigation solution. The algorithm has been implemented and validated based on the data obtained by the AIMS system, originally developed at the Center for Mapping, The Ohio State University. AIMS utilizes the DD GPS carrier phase observable as the measurement update (and requires integer ambiguity values), while the method presented here is based on introducing the GPS velocity derived from the carrier phase observable to bridge these portions of the trajectory, where GPS signal is partially or entirely available but the integer ambiguities cannot be established. The preliminary results indicate that a steady navigation accuracy can be maintained for the entire trajectory. Moreover, the GPS velocity information, combined with the DD carrier phase, is likely to improve the filter's sensitivity to the platform maneuvers.